Impact of supplementing newborn infants with vitamin A on early infant mortality: community based randomised trial in southern IndiaBMJ 2003; 327 doi: https://doi.org/10.1136/bmj.327.7409.254 (Published 31 July 2003) Cite this as: BMJ 2003;327:254
- Lakshmi Rahmathullah, director1,
- James M Tielsch, professor ()2,
- R D Thulasiraj, executive director3,
- Joanne Katz, professor2,
- Christian Coles, assistant research professor2,
- Sheela Devi, research coordinator3,
- Rajeesh John, biostatistician3,
- Karthik Prakash, biostatistician3,
- A V Sadanand, faculty3,
- N Edwin, professor4,
- C Kamaraj, professor4
- 1Aravind Centre for Women, Children and Community Health, Madurai, Tamil Nadu, India
- 2Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205-2103, USA
- 3Lions-Aravind Institute for Community Ophthalmology, Madurai, Tamil Nadu, India
- 4Department of Pediatrics, Madurai Medical College, Madurai, Tamil Nadu, India
- Correspondence to: J M Tielsch
- Accepted 16 April 2003
Objective To assess the impact of supplementing newborn infants with vitamin A on mortality at age 6 months.
Design Community based, randomised, double blind, placebo controlled trial.
Setting Two rural districts of Tamil Nadu, southern India.
Participants 11 619 newborn infants allocated 24 000 IU oral vitamin A or placebo on days 1 and 2 after delivery.
Main outcome measure Primary outcome measure was mortality at age 6 months.
Results Infants in the vitamin A group had a 22% reduction in total mortality (95%confidence interval 4% to 37%) compared with those in the placebo group. Vitamin A had an impact on mortality between two weeks and three months after treatment, with no additional impact after three months.
Conclusion Supplementing newborn infants with vitamin A can significantly reduce early infant mortality.
Ocular signs of vitamin A deficiency are associated with increased mortality among children aged 6 months or older.1 Supplementation with vitamin A can significantly reduce total mortality.2–8 The impact of supplementation is only clear in children aged 6 months or older. It was assumed that breast feeding protected infants from vitamin A deficiency and that clinical signs of xerophthalmia appeared only after the onset of weaning or incorrect bottle feeding.1 Recent evidence challenges these assumptions. Infants are born withlow stores of vitamin A and depend on feeding to build body stores.9–11 If the mother's breast milk has a low concentration of vitamin A, as found in many women in developing countries, the infant will be unable to meet daily requirements and increase body reserves.12
Supplementation of women post partum can improve the concentration of vitamin A in their blood and breast milk, but data on the impact of supplementing infants early with vitamin A on survival are conflicting.13 14 A study in Nepal found no difference in mortality between infants supplemented with vitamin A orplacebo before 6 months of age (relative risk 1.11, 95% confidence interval 0.86 to 1.42).15 A trial conducted in three countries also found no impact on mortality in the first six months from supplementing with vitamin A both mothers post partum and their infants at all three visits for diphtheria, tetanus, and pertussis immunisations.16 Another study in Nepal found no effect on early infant mortality from supplementing women with vitamin A or β carotene before and during pregnancy.17 In contrast, supplementing infants with 50 000 IU of vitamin A within 24 hours of birth was safe and associated with a 64% reduction in infant mortality.18 We aimed to determine the impact of supplementing infants with vitamin A within 48 hours of delivery on early infant mortality.
Our study (the vitamin A supplementation in newborns (VASIN)) was a randomised, placebo controlled, community based trial conducted between June 1998 and March 2001 in two rural districts of Tamil Nadu, southern India.
Eligibility and randomisation
Liveborn infants that resulted from all pregnancies within participating villages were eligiblefor participation. Pregnant women were identified for recruitment from a variety of sources. Project staff explained the study to them and attempted to recruit them before delivery. Verbal informed consent was obtained at the time of recruitment. Baseline information on personal and socioeconomic characteristics was collected by interview. Pregnant women in this area of India traditionallymove in with their parents for delivery and for up to six months after delivery. Eligibility wastherefore determined by where the woman delivered her child.
Randomisation was at the individual level, stratified by geographical area in blocks of four. Because births were likely in a variety of locations, randomisation was conducted at the time of recruitment. Exclusions after randomisation were stillbirths, miscarriages, delivery more than 20 kmoutside the study area, and infants who died before our study team arrived. For twins, the first born received the assigned treatment and the second born the other treatment. For triplets, the first two born infants were handled as twins and the third born received the originally assigned treatment.
Intervention and data collection
The infants were randomly assigned to receive either 24 000 IU of vitamin A twice within a 24 hour interval, beginning within 48 hours of birth, or placebo. The treatment doses were in an edible oil solution packaged in identical gelatin capsules. Mothers were encouraged to breast feed their infant immediately after treatment to ensure consumption of the full dose. Each set of two capsules was packaged separately and labelled with the mother's identifying information. Investigators,study staff, and mothers were masked to the assigned treatment. Treatment codes were kept in a sealed envelope in a locked filing cabinet in Baltimore.
Village based staff notified their supervisor when a birth had occurred. The supervisor travelled to the site of the delivery to provide the assigned treatment, weigh the infant, and collect information on the delivery. Supervisors had a target to begin treatment within 48 hours of birth oras soon as possible if this was not achievable. Weight was measured with an electronic infant weighing scale (Seca Model 727; Seca, Columbia, MD).
The day after the first dose, the supervisor revisited the household to collect any comments ontreatment and to provide the second dose. In the case of adverse events, a report was completed and the child visited daily for seven days.
Project staff visited the household every two weeks to assess the vital status of the child andany morbidity. Infants were followed until 6 months of age. Before discharge the infants had anthropometric measurements taken and were given a 100 000 IU dose of vitamin A. We considered as censored those infants aged less than 6 months who were being followed at the end of March 2001.
Measurement of outcomes
The primary outcome was mortality within the first six months of life. Infant deaths were ascertained during the vital status and morbidity assessments. Cause of death was determined from an independent review of verbal autopsy information from family members by two paediatricians. Disagreements were resolved by consensus with a third paediatrician.
Sample size and statistical analysis
Our study was designed to detect a minimum reduction in infant mortality at age 6 months of 30%; given an expected infant mortality at 6 months of age of 52.5 per 1000 live births, 80% power, a 5% two sided type I error, and 10% loss to follow up. We required 3000 live births per group. Unknown was the number of deaths that would occur before infants were enrolled into our study. An independent data and safety monitoring board, which met twice to review our data, advised us to increase the recruitment period to be able to detect a 20% reduction among enrolled infants because the mortality of enrolled infants was lower than expected due to early deaths. We therefore required 4500 live births per group.
Statistical analysis was performed with SAS. Treatment groups were compared on baseline household, maternal, and infant characteristics for all deliveries and for liveborn infants who were enrolled. We used three approaches for the primary analysis of treatment effect on mortality. Firstly, we estimated the incidence density of mortality with person time as the denominator. This permitted use of all data, including infants who were censored at the time follow up was completed. Secondly, we estimated the infant mortality at age 6 months, with live births as the denominator. We included in this analysis only infants who were followed to 6 months. Thirdly, we performed a survival analysis. Cox proportional hazard models were used to adjust for potential confounding and to model potential effect modification. All analyses were based on intention to treat.
We recruited 14 035 pregnant women; 862 of these did not deliver within the study area (fig 1).Overall, 13 294 infants were born; 6670 (50.2%) were allocated placebo and 6624 (49.8%) were allocated vitamin A. The parents of 23 infants refused to participate after delivery, families of 1027 infants moved out of the area, and 625 were stillbornor died before our teams arrived. This left 11 619 liveborn infants enrolled and followed; 5833 (50.2%) in the placebo group and 5786 (49.8%) in the vitamin A group.
Baseline characteristics of the families, mothers, and infants were similar between the treatment groups (tables 1, 2, 3, 2). Thisapplied to all deliveries and to those infants who were enrolled. About 13% of mothers were aged under 20 at the time of the pregnancy. Over 40% of mothers had no formal education. About 8% reported a history of miscarriage, and 4% reported a previous stillbirth. Over 60% of births were attended by a physician or nurse, and around 20% were attended by a trained midwife. Five to 6% of womenreported a history of night blindness, a clinical symptom of vitamin A deficiency. The mean birth weights were 2675 g in the placebo group and 2673 g in the vitamin A group (31% of infants weighedless than 2500 g). Similar proportions of mothers in both groups began breast feeding within 12hours of birth and reported feeding colostrum. Mortality of infants born alive but not enrolled was similar in both groups (2). Eighty per cent of infants werefirst dosed within 48 hours of birth (median time to first dose, 25.5 hours in placebo group and 26.4hours in vitamin A group).
Side effects were uncommon. Six cases occurred in the placebo group and three in the vitamin A group. Most of these were vomiting; there were no reports of bulging fontanelle.
Supplementing newborn infants with vitamin A was associated with a 22-23% reduction in mortality during the first six months of life (table 4 and 3).Similar estimates of relative risk were obtained with the incidence density analysis (0.78, 95% confidence interval 0.63 to 0.96), infant mortality at 6 months(0.77, 0.62 to 0.96), and an estimate of the hazard ratio (hazard ratio 0.78, 0.63 to 0.97). No substantial change was observed in these estimates when adjusting for additional baseline variables (data not shown). The survival curves began to diverge at around 2 weeks of age and continued to separate until 3 months of age (fig 3). After 3 months the curvesremained parallel, indicating no further treatment effect.
Little evidence was found for an effect modification by sex (table 4, test for interaction, P=0.33). The impact of vitamin A on survival was limited to infants who were treated before 14 days (table 5), although the strength of evidence for this interaction was low (test for interaction, P=0.68). The effect was also limited to infants of low birth weight (table 6). In this group, vitamin A reduced mortality at 6 months by 37%. In contrast, there was no effect of vitamin A on mortality among infants weighing 2500 g or more at birth (test for interaction, P=0.02).
Giving newborn infants two doses of 24 000 IU of vitamin A within 48 hours of birth significantly reduced early infant mortality. Our results agree with those from a hospital based study in Indonesia, which reported a 64% reduction in infant mortality associated with giving newborn infants 50 000 IU of vitamin A.18 As in our results, the impact of vitamin A on survival was limited to the first three or four months of life. The greatest impact in the Indonesian study was among infants who weighed 2500 g or more at birth. In contrast, we found the greatest effect among infants of low birth weight.
The discrepancy with studies that supplemented infants later in the first six months suggests something unique about receiving a large dose of vitamin A shortly after birth.15–17Although the underlying mechanism for this differential impact by age is unknown, two explanations are plausible. Humans are born with marginal reserves of vitamin A and depend on breast milk or other sources to meet their metabolic demands in the first few months of life. Premature infants have even lower reserves of vitamin A, and correction of the deficiency has been shown to reduce the respiratory complications of preterm birth.19 A large bolus of vitamin A early in the neonatal period may provide a stimulus to rapid maturation of both gut and respiratory epithelium. This matured epithelium may be more resistant to invasion by pathogens or may clear such organisms more efficiently.
What is already known on this topic
Supplementation with a large dose of vitamin A reduces mortality among children aged 6 months to 5 years in many developing countries
The effect of supplementation on mortality in newborn infants under 6 months of age is unclear
Periodic treatment with vitamin A beyond the first month of life has no impact on mortality
What this study adds
Supplementing infants with vitamin A in the first few days after birth significantly reduced infant mortality
The greatest impact was observed between 2 weeks and 3 to 4 months of age
Another potential mechanism relates to the role of vitamin A in the development and maintenanceof immunocompetence. Vitamin A deficiency causes alterations in T cell subsets, impaired phagocytic activity, and reduced antibody response to antigen challenge.20–22Retinoic acid is an important regulator of gene expression and cell differentiation, and vitamin A deficiency can cause alterations in the balance of Th1 versus Th2 type cytokine response.23 24However, these effects have been observed in animal models and humans after the newborn period.
Our study had limitations. The need to maintain control over delivery of the assigned treatment, and the variation in locations and time at which deliveries took place, resulted in a delay in arrival at the place of delivery. We therefore missed several children who were born alive and subsequently moved, whose parents refused participation, or who died before the arrival of our staff. Although there was no difference in survival between the treatment groups before enrolment, the delay prevented estimation of the effect of supplementation on total six month infant mortality.
A table on cause specific mortality appears on bmj.com
We thank G Venkataswamy and P Namperumalswamy for their support, the members of the Data and Safety Monitoring Board, including the late V Ramalingaswamy (All India Institute of Medical Sciences, New Delhi), P S S Sundar Rao (Schieffelin Leprosy Research and Training Centre, Karigiri), S J M Jeyam (Institute for Child Health, Chennai), K Vijayaraghavan (National Institute of Nutrition, Hyderabad), and Vijaya Srinivasan (Gandhigram Institute of Rural Health and Family Welfare).
Contributors LR, JT, RDT, and JK helped design the study. LR, JT, RDT, CC, SD, and AVS supervised the conduct of the study, JT, RDT, JK, RJ, and KP were involved in the analysis and interpretation of the data. LR, CC, NE, and CK were involved in the interpretation of the data. JK and KP designed the data management system. NE and CK reviewed the verbal autopsies and determined the cause of death for the study. All authors contributed to writing or editing the manuscript, or both. JT will act as guarantor for the paper.
Funding This study received support from Cooperative Agreement No HRN-A-00-97-00015-00 between the Center for Human Nutrition (Bloomberg School of Public Health, Johns Hopkins University) and the Office of Health and Nutrition (US Agency for International Development, Washington DC), the Bill and Melinda Gates Foundation (Seattle, Washington), and Task Force Sight and Life (Basel, Switzerland). These financial supporters had no role in the design, conduct, analysis, or reporting of this study.
Competing interests JMT, JK, and CC have received support from the Sight and Life Institute, Center for Human Nutrition, Johns Hopkins Bloomberg School of Public Health. The institute was established with a gift from Task Force Sight and Life, Roche.
Ethical approval: This study was approved by the Aravind Eye and Children's Hospitals, the Department of Health, Tamil Nadu state government, and the Committee on Human Research of the Johns Hopkins Bloomberg School of Public Health.